The role of reinforcement in affective computation

In this paper we review the functional role that reinforcement plays in notions of affective computation and emotion. We consider three core components of emotional activity - emotion triggers, actions and action tendencies, and feeling states - and evaluate each component in relation to reinforcement learning and behaviour theory as well as with respect to existing neural-computational models. We propose that although such a categorization of emotional components can be helpful to understanding affective computation and the role reinforcement plays, the three components are dynamically inter-dependent grounded upon reinforcement learning mechanisms fundamental to survival.